This message appears when the Solver could not find any combination of values for the decision variables that allows all of the constraints to be satisfied simultaneously. If you are using the LP/Quadratic Solver or the SOCP Barrier Solver, and the model is well scaled, the Solver has determined for certain that there is no feasible solution.
If you are using the nonlinear GRG Solver, the GRG method (which always starts from the initial values of the variables) was unable to find a feasible solution; but there could be a feasible solution far away from these initial values, which the Solver might find if you run it with different initial values for the variables.
If you are using the Interval Global Solver, this message means that the Solver could find no feasible solutions after a systematic exploration of the search space. The Interval Global Solver is designed to “prove feasibility” as well as global optimality, and there is very likely no feasible solution; but this is subject to limitations due to the finite precision of computer arithmetic – discussed below in “Limitations on Global Optimization” – that can, in rare cases, cause the Solver to “miss” a solution.
If you are using the Evolutionary Solver, the evolutionary algorithm was unable to find a feasible solution; it might succeed in finding one if you run it with different initial values for the variables and/or increase the Precision value on the Task Pane Engine tab (which reduces the infeasibility penalty, thereby allowing the evolutionary algorithm to explore more “nearly feasible” points).
If you are solving a problem with chance constraints using simulation optimization, this message means that the Solver could find no solution that satisfies these constraints to the chance measures (such as 95%) that you specified. If you ‘relax’ the chance measures (to say 90%) and solve again, it’s possible that a feasible solution will be found. For robust optimization, see result codes 26 through 29.
In any case, you should first look for conflicting constraints, i.e. conditions that cannot be satisfied simultaneously. Most often this is due to choosing the wrong relation (e.g. <= instead of >=) on an otherwise appropriate constraint. The easiest way to do this is to select the Feasibility Report, shown in the Reports list box when this message appears, and click OK. (This report can take time for the LP/Quadratic Solver or SOCP Barrier Solver, more time for the GRG Solver, and much more time for the Interval Global Solver; it is not available for the Evolutionary Solver.) For an example of using the Feasibility Report to diagnose an infeasible solution, see “The Feasibility Report” in the chapter “Solver Reports.”